By JayR Cela 5/3/08
Most SLifers are aware of Linden Labs(tm) fervently fast paced upgrade path for the SecondLife(tm) client viewer. The current batch, 1.20.x.(1-5) have been met with mixed results from the general and developer communities. During a recent Linden meeting, I had the chance to ask some questions, and print a condensed version of my transcript. But first A quick look at the disappearing skirts issue, the Havok 4 team, along with some closing comments on recent Server problems, and a statement from Rob Linden.
Disappearing Skirts
You: silly question / if I may ask please ...............
You: has any one solved the dissapearing skirt problems yet ?
Alexa Linden: it's still being looked at, VWR-2778
You: ok / thanks Alexa :_)
Alexa Linden: np :), it's been imported, and we are looking at it
The Havok 4 Team
You: ok i got comment.......
Aric Linden listens to JayR
You: sidewinder and his team deserve some real hero pats on the back and probably a raise in pay for pulling of the havok 4 release, that team rocks
Aric Linden: Oh, they got all that and more, and yes they do rock. They're showing the rest of us how real people get work done
Dazzle and the 1.20.x.x Release Candidates
You: so why ya'all gonna do about Dazzle ? / I personally like it / but seems as
if not many other folks do ?
Aric Linden: JayR? what is it that you think we should do with Dazzle?
You: ok / I personally like the Dazzle skin / should be an option to choose between
Classic and Modern / same as Sea Monkey ( FireFox ) (Win~Amp ) why is this
not an option ?
Kooky Jetaime: I haven't tried Dazzle... I'm in the boat though of hold dazzle and
just go straight to a skinnable UI. At the minimum, give the choice of Dazzle or
raditional. I like the dark ui.
Aric Linden: It's not an option currently for a few reasons. It was very expensive to try and test and support two versions
concurrently so we opted for pushing to dazzle and moving towards a skinnable ui
Squirrel Wood: Dazzle, afaik is a first step towards skin ability
Aric Linden nods at Squirrel
Yuu Nakamichi: I have read that the skinnaibility will be finishied before Dazzle
becomes main?
Asriazh Frye: would the skin modifying be an in-client thing or an extra program?
Aric Linden: Asriazh it would be a simple config - you'd add a folder with your
desired skin to the install directory for SL. We are working on it, and we're split into many different groups and have a range of priorities. Sometimes our priorities map 1-1 with residents (the crashes in last weeks rc for example). In other cases, the timing shifts out based on other ordering for resources. QA and release are both stretched very tightly right now, we wanted to get Dazzle out and get feedback before we completed the skinning, but we haven't given up on skinning at all .
You: Aric / seems as if QA is streatched more than a bit thin / the RC's should be
re-named ALPHA clients / they are not of Release Canidate calibur
Aric Linden laughs This topic has been discussed and is in the process of being worked out now
I'm driving that
You: and this is a problem ...... because ..... the average user is or does not
seem to be able to grasp the fact that these are experimental
Squirrel Wood: mayhaps a name shift from RC to Beta Client may help ?
Aric Linden: but since you seem to know - how would you define the difference
between Release candidate and Alpha or Beta software
You: Ok...look at MS for an example, they go thru and closed alpha & beta stage .....
Aric Linden thinks fondly of IE 3 and listens closely
You: then an open beta.....then a closed RC....then an open RC stage
Aric Linden: and they take how long to release product?
You: a long time:_)
Aric Linden: would you be willing to wait 2 years for an update to SL?
You: no / I would be willing to wait several months. What the heck are you'all rushing for ?
Aric Linden: We're rushing because we are not ubiquitous. We are trying to build a platform and feel that time to market matters. We are also solving (or trying to solve) problems that MS doesn't have
Trinity Coulter: for me, i don't think rapid changes matter as long as you have a
lot of communication with the residents
Aric Linden: all that said, we will be slowing down the RC process and implementing
a more closed alpha cycle
You: but in the mean time you have created a unstable platform ..... why not pay residents that actually want to work with you..... for their time to test things before they are release into the general population
Aric Linden: Um, I'm not in charge of the pay residents to work with us problem, JayR - I will bring your suggestion to the team tho
Kooky Jetaime: I think JayR is on something that I've been repeating about for awhile....
LL needs to use the "established" meanings for words in software development.
These RC's are not Release Candidates. They are Public Beta's. They are no where NEAR polished, or feature locked
Aric Linden: Kooky - have you been sitting in on our meetings?
Kooky Jetaime: I hit some hours here and there, if not talking about those, then no.
Aric Linden: no, i meant our internal meetings
Kooky Jetaime: No.
Aric Linden: we're discussing this stuff - alpha beta naming conventions time to
release etc.. this is a big topic internally and we'll be making some changes
Bridie Linden: Another reason for the increase in RCs lately was to try and
find/fix crashes...
You: but in essance you are using the general Avatar population as guinee piggs.....
thus in the long run ...alienaiting a lot of potential future customers to this platform
Kooky Jetaime: Bridie - RC's are not for bug hunting. Thats what Alphas/Betas are for....
something titled with "RC" should be, as far as can be determined" A-OK, and ready to
ship, your just checking to make sure the public agrees with you.
Bridie Linden: As Aric mentioned, we are discussing all of this internally.
Aric Linden: thanks - we're doing something that's not been done before and we're
feeling our way along sometimes really painfully, sometimes really successfully. We're' trying to collaborate with you all and internally
Bridie Linden nods
Aric Linden: sometimes we don't realize we've effed up until two days after the effup
Squirrel Wood: If y'all ask me then it may be a wise idea to have one team rewrite
the client from scratch and define common interfaces for different modules like input,
rendering, skins, ... but that's just my two cents of L$
You: i agree with Squirrel
Aric Linden: wow squirrel - i think if we had a large enough team, we'd consider that, but that's HUGE, and by the time we got to the end of that, we'd have to rewrite a
bunch of it cuz it was outdated
Trinity Coulter: they are working toward that, Squirrel, by changing the way the server and client interact, so its on the long range plan
Squirrel Wood: yes. but the team would just rebuild the base framework. everything
else comes as module to link into that frame
Aric Linden: Trinity is right, exactly right, and even though sometimes it doesn't seem like it, we try to factor residents into the work we do all the time
Some Server Issue's
You: personally I dont think the client is the biggest problem right now
Aric Linden: agreed jayr, the server is. But that's only because we put so much into the client over the last 9 months
Squirrel Wood: Yes. The asset server has been a pain during the past couple weeks...
not a single day without any major troubles
Aric Linden: we're now pointing at trying to get the server solid
Trinity Coulter: to help improve the reliability of SL, but, i guess this isn't the Which hour, its the Bug Lindens hour :)
Squirrel Wood: modularize...much easier to rewrite or exchange a module than the entire thing :)
Aric Linden: Squirrel, we're trying, we're getting there
You: Backward compatibility with legacy code is allways a problem / and sometimes,
can choke off inovation /
Aric Linden:sometimes, so you refactor all of that and make occasional updates mandatory, and then listen to your users gripe about mandatory updates, or lack of mandatory updates
Aric Linden: Sadly, I need to go, I'll set up office hours. It seems as though this is a conversation worth continuing with y'all
Final Comment from Rob Linden
Rob Linden: the question is not whether it's possible to write something
more compact and less buggy. the question is whether a complete rewrite is a faster path to
more compact and less buggy than gradual refactoring/improvement
Rob Linden: this line taken from the essay link below, is pretty important:
http://www.joelonsoftware.com/articles/fog0000000069.html
" The old mantra build one to throw away is dangerous when applied to large scale commercial
applications. "
Well there ya have it folks.
JayR Cela :_)
No comments:
Post a Comment